Trying to Generate Random Data from Lists - r

Question

I am trying to generate data for a project. The data needs to be generated randomly from predefined lists. Essentially, I have real data but it's very small. In order to build some classifiers (decision tress, Support Vector Machines and Naive Bayes), I want to produce 100,000 observations.
I am new to coding (I can do rudimentary things in Matlab and R) and initially tried to do this in Excel, however, the RANDOMA function generated very equally distributed data. To be more specific, I am using 5 demographic pieces of information to predict which retailer a customer will select, e.g. retailer A, B or C. The lists for the demographic information is below:
1) Age group (18-24, 25-34, 35-44, 45-54, 55+)
2) Gender (male or female)
3) Income group (<£10k,£10k-19.99k, £20k-£29.99k, etc.)
4) Region (London, Wales, Scotland, Nothern Ireland, South West, etc.)
5) Type of job (Full-time, part-time, student, etc.)
When I tried to randomly create 100,000 observations (each observation randomly selected 1 from each of the 5 lists), they were almost equally distributed between them. Even worse, the value you I randomly assigned to the retailer (A, B or C) was also equal.
The idea is to split this randomly generated data into training and test data, so I can build some models and test their suitability.

Answer 1

In Matlab, your best friend for this task will be randsample function (reference here), which is part of the Statistics Toolbox. Let's make an example concerning your Gender variable:
% possible values (M for male and F for female)
% since it's a qualitative variable, let's use the categorical type
var = categorical({'M' 'F'});
prob = [0.55 0.45]; % corresponding probabilities
n = 100000; % sample size
repl = true; % replacement (true = yes, false = no)
gender = randsample(var,100000,repl,prob);
You can use the same approach to generate samples concerning Region and Job. Let's now make another example with your Age variable.
var = 1:100; % possible values (age from 1 to 100 years)
n = 100000; % sample size
repl = true; % replacement (true = yes, false = no)
% the probability argument is not provided, hence the result is equally distributed
age = randsample(var,100000,repl);
Since you want to split your Age sample into different groups, the histcounts with edges as the second argument will do that for you:
age_grps = histcounts(age,[0 18 25 35 45 55 100]);
% remove the first column if you want to esclude people from 0 to 17 years
age_grps(1) = [];
You can use the same approach to generate the Income sample.
As far as I can see, your main concern is the uniform distribution of your variables. I show you how to set different probabilities for each possible value in the randsample function (prob argument).

Answer 2

I don't know the typical distributions of your data, but the following should get you started.
library(tidyverse)
set.seed(315) # This will create the same data set each run
n.size <- 500
myData <- tibble(
ID = 1:n.size,
VisitDT = lubridate::today()-30 - (runif(n.size) * 100),
IncomeGroup = sample(c("Low", "Medium", "High" ), n.size, prob = c(.7, .25, .05), replace = TRUE),
age = round(rnorm(n = n.size, mean = 52, sd = 10),2),
sex = sample (c('M', 'F'), size = n.size, prob = c(.4, .6), replace = TRUE),
region = sample (c('London', 'Wales', 'Scotland'), size = n.size, prob = c(.4,.3,.2), replace = TRUE),
Treatment = sample(c('No','Yes'), size = n.size, prob = c(.1, .9), replace = TRUE)
)